Reviewer's comment Hello, my name is Sveta Nosova and I am going to review this project. Before we start, I want to pay your attention to the color marking:

Reviewer's comment Great solutions and ideas that can and should be used in the future are in green comments.

Reviewer's comment Yellow color indicates what should be optimized. This is not necessary, but it will be great if you make changes to this project.

Reviewer's comment Issues that need to be corrected to get right results are indicated in red comments. Note that the project cannot be accepted until these issues are resolved.

---

Please, use some color other than those listed to highlight answers to my comments. I would also ask you not to change, move or delete my comments so that it would be easier for me to navigate during the next review.

In addition, my comments are defined as headings. They can mess up the content, however, they are convenient, since you can immediately go to them. I will remove the headings from my comments in the next review.

---

**A few words about the project:** you did a great job, everything is clear and neat, there are analysis and intermediate conclusions. The project has a lot of graphs, which is great. I still have some questions that I've written in my comments. I've also left there some recommendations for improving the project. If you hava any questions, feel free to ask me.

---

**About the presentation:**

• PDF is covenient and portable, right choice. That's why it's recommended to use this format.

• Technical description is an important part. It makes the report more presentable.

• It's good to use a white background instead of a colored one. We never know, if we have to show the presentation on a projector, where the colored background can spoil the perception, while everything is always clearly visible on white background.

• It's also very good that you use the main idea of a slide as a title.

• I see a slight distortion on the graphs. Try saving them using fig.savefig('fig.png', dpi=150) with dpi parameter that represents a resolution in dots per inch.

• It's highly recommended to use no more than one graph on each slide, so everything is clearly visible.

• Do not forget to add slide numbers on your presentations. It will be easier for the reader to mark something on slide and easily get back to it.

• The slide with conclusions and recommendations can also be located at the end of the presentation. It's not a strict rule to place it in the beginning. So, you can place it at the end too.

</div>

Student's comment: Hi Sveta, Thank you for you reviewing my project. I want to be a good coder, so I appreaciate your feedback! I've made updates below. Thank you, Jolene G.

Reviewer's comment 2 Jolene, thank you for sending the corrected version of the project. You've done a fantastic job 😊 I've left a few comments titled as **Reviewer's comment 2**. I've also removed the headings from the old comments. Hope, my comments were helpful. Your project has passed code review. Congratulations 😊 Take care and good luck!

---

Best regards, Sveta

Telling a Story with Data Course Project

Project Description

You’ve decided to open a small robot-run cafe in Los Angeles. The project is promising but expensive, so you and your partners decide to try to attract investors. They’re interested in the current market conditions — will you be able to maintain your success when the novelty of robot waiters wears off?

Tasks

• You’re an analytics guru, so your partners have asked you to prepare some market research.
• You have open-source data on restaurants in LA.

Instructions for completing the project

Step 1. Download the data and prepare it for analysis

• Download the data on restaurants in LA
• Evaluate the data
  • Make sure that the data type for each column is correct
  • that there are no missing values or
  • that there are no duplicates
    • Process them if necessary

Step 2. Data analysis: Perform the following tasks:

• Investigate the proportions of the various types of establishments. Plot a graph.
• Investigate the proportions of chain and nonchain establishments. Plot a graph.
• Which type of establishment is typically a chain?
• What characterizes chains: many establishments with a small number of seats or a few establishments with a lot of seats?
• Determine the average number of seats for each type of restaurant. On average, which type of restaurant has the greatest number of seats? Plot graphs.
• Put the data on street names from the address column in a separate column.
• Plot a graph of the top ten streets by number of restaurants.
• Find the number of streets that only have one restaurant.
• For streets with a lot of restaurants, look at the distribution of the number of seats. What trends can you see?

Draw Conclusions

• Draw an overall conclusion and provide recommendations on restaurant type and number of seats. Comment on the possibility of developing a chain.

Data Description

• Use the following data: /datasets/rest_data_us.csv
  • object_name — establishment name
  • chain — chain establishment (TRUE/FALSE)
  • object_type — establishment type
  • address — address
  • number — number of seats

• Step 3. Preparing a Presentation
  • Make a presentation of your research to share with investors.
    • It must be converted to PDF format for assessment
    • use the following format in a markdown cell: < Presentation: >

Plan of Action

1. Import Libraries
2. Open the data files and study the general information (original dataframe)
3. Data Preprocessing (working dataframe)
   • Create working versions of the dataframes
   • Clean up the data
4. Analyze the data
   • Answer the requested information
5. Draw conclusions
6. Prepare a presentation

Reviewer's comment There's an introduction, which is good. It is important to write an introductory part, because it gives an idea about the content of the project.

1. Import Libraries

import pandas as pd
import numpy as np
import matplotlib.pyplot as plt
import datetime
import seaborn as sns
from scipy import stats as st
from plotly import graph_objects as go

import sys
import warnings
if not sys.warnoptions:
    warnings.simplefilter("ignore")
    
sns.set_palette('bright')

2. Open the data files and study the general information (original dataframe)

Read data as a dataframe and examine original data

# create dataframe to examine original data
cafe_data_orig = pd.read_csv('datasets/rest_data_us.csv')

Examine original df

cafe_data_orig.info()

<class 'pandas.core.frame.DataFrame'>
RangeIndex: 9651 entries, 0 to 9650
Data columns (total 6 columns):
 #   Column       Non-Null Count  Dtype 
---  ------       --------------  ----- 
 0   id           9651 non-null   int64 
 1   object_name  9651 non-null   object
 2   address      9651 non-null   object
 3   chain        9648 non-null   object
 4   object_type  9651 non-null   object
 5   number       9651 non-null   int64 
dtypes: int64(2), object(4)
memory usage: 452.5+ KB

cafe_data_orig.head()

	id	object_name	address	chain	object_type	number
0	11786	HABITAT COFFEE SHOP	3708 N EAGLE ROCK BLVD	False	Cafe	26
1	11787	REILLY'S	100 WORLD WAY # 120	False	Restaurant	9
2	11788	STREET CHURROS	6801 HOLLYWOOD BLVD # 253	False	Fast Food	20
3	11789	TRINITI ECHO PARK	1814 W SUNSET BLVD	False	Restaurant	22
4	11790	POLLEN	2100 ECHO PARK AVE	False	Restaurant	20

cafe_data_orig['id'].nunique()

9651

cafe_data_orig['object_name'].nunique()

8672

cafe_data_orig['address'].nunique()

8517

cafe_data_orig['chain'].value_counts(dropna = False)

False    5972
True     3676
NaN         3
Name: chain, dtype: int64

null_query = cafe_data_orig.loc[cafe_data_orig['chain'].isnull()]
null_query

	id	object_name	address	chain	object_type	number
7408	19194	TAQUERIA LOS 3 CARNALES	5000 E WHITTIER BLVD	NaN	Restaurant	14
7523	19309	JAMMIN JIMMY'S PIZZA	1641 FIRESTONE BLVD	NaN	Pizza	1
8648	20434	THE LEXINGTON THEATER	129 E 3RD ST	NaN	Restaurant	35

search_1 = cafe_data_orig.query('object_name == "TAQUERIA LOS 3 CARNALES"')
search_2 = cafe_data_orig.query('object_name == "JAMMIN JIMMY\'S PIZZA"')
search_3 = cafe_data_orig.query('object_name == "THE LEXINGTON THEATER"')
display(search_1)
display(search_2)
display(search_3)

	id	object_name	address	chain	object_type	number
7408	19194	TAQUERIA LOS 3 CARNALES	5000 E WHITTIER BLVD	NaN	Restaurant	14

	id	object_name	address	chain	object_type	number
7523	19309	JAMMIN JIMMY'S PIZZA	1641 FIRESTONE BLVD	NaN	Pizza	1

	id	object_name	address	chain	object_type	number
8648	20434	THE LEXINGTON THEATER	129 E 3RD ST	NaN	Restaurant	35

cafe_data_orig['object_type'].value_counts(dropna = False)

Restaurant    7255
Fast Food     1066
Cafe           435
Pizza          320
Bar            292
Bakery         283
Name: object_type, dtype: int64

2 Conclusion of original dataframe and general information

Actions performed

• Examined original data for general information
  • the dataset was not modified in this section

Dataframe info

• there are 9651 entries each with a unique id
• there are 8672 unique establishment names ('object_name')
  • suggesting there are chains or duplicates
• there are 8517 unique addresses
  • this suggests that there are duplicates or multiple locations at one address, for example at a mall or shopping center
• 5972 are not chains, 3676 are chains, and 3 are unknown
  • the 3 unknown do not have any other names in the list, suggesting they are not a chain
• there are 6 different types of establishments ('object_type)
  • most are a 'Restaurant' type

Reviewer's comment Nice intro.

3 Data Preprocessing

Clean up the data and process where necessary

Plan of Action

• read and save working versions of data
• fill in nulls as 'False' since no other locations with that name were found
• optimize datatype storage
• retitle columns to be more readable/user friendly
• identify missing values
  • examine and process
• identify duplicate rows
  • examine and process

Due to the low numbers

Save working version of the data

# create working versions of the dataframes for data analysis
cafe_data = pd.read_csv('datasets/rest_data_us.csv')

Fill null values

• null chain values, were examined and found to not have mulitple establishments - this suggests that it is not a chain
• these values will be filled with False, i.e. they are not a chain

cafe_data.loc[cafe_data['chain'].isnull(), 'chain'] = False
cafe_data.info()

<class 'pandas.core.frame.DataFrame'>
RangeIndex: 9651 entries, 0 to 9650
Data columns (total 6 columns):
 #   Column       Non-Null Count  Dtype 
---  ------       --------------  ----- 
 0   id           9651 non-null   int64 
 1   object_name  9651 non-null   object
 2   address      9651 non-null   object
 3   chain        9651 non-null   object
 4   object_type  9651 non-null   object
 5   number       9651 non-null   int64 
dtypes: int64(2), object(4)
memory usage: 452.5+ KB

Optimize datatype storage

# optimize datatype storage
cafe_data['id'] = pd.to_numeric(cafe_data['id'], downcast = 'integer')
cafe_data['object_name'] = cafe_data['object_name'].astype('category')
cafe_data['address'] = cafe_data['address'].astype('category')
cafe_data['chain'] = cafe_data['chain'].astype('bool')
cafe_data['object_type'] = cafe_data['object_type'].astype('category')
cafe_data['number'] = pd.to_numeric(cafe_data['number'], downcast = 'integer')

cafe_data.info()

<class 'pandas.core.frame.DataFrame'>
RangeIndex: 9651 entries, 0 to 9650
Data columns (total 6 columns):
 #   Column       Non-Null Count  Dtype   
---  ------       --------------  -----   
 0   id           9651 non-null   int16   
 1   object_name  9651 non-null   category
 2   address      9651 non-null   category
 3   chain        9651 non-null   bool    
 4   object_type  9651 non-null   category
 5   number       9651 non-null   int16   
dtypes: bool(1), category(3), int16(2)
memory usage: 745.0 KB

Retitle columns to be more readable/user friendly

# review original data column titles
cafe_data.columns

Index(['id', 'object_name', 'address', 'chain', 'object_type', 'number'], dtype='object')

# retitle columns to be readable/user friendly
cafe_data = cafe_data.rename(columns = {'id': 'location_id', 'object_name': 'location_name', 'object_type': 'location_type', 'number': 'num_seats'})
# review updated column titles
cafe_data.columns

Index(['location_id', 'location_name', 'address', 'chain', 'location_type',
       'num_seats'],
      dtype='object')

Identify missing values in the dataset

print('number of missing values in df')
cafe_data.isnull().sum()

number of missing values in df

location_id      0
location_name    0
address          0
chain            0
location_type    0
num_seats        0
dtype: int64

Identify duplicate values in the dataset

duplicate_rows = cafe_data.duplicated().sum()
print('number of duplicate rows: {}'.format(duplicate_rows))

number of duplicate rows: 0

• When examining the original data, 8517 unique addresses were found out of a data set with 9651 entries, suggesting that there may be duplicate entries of the same location and each were given a unique 'id'
  • Evaluate the loacation name and address for duplicates
  • Evaluate the addess duplicates

# determine if there are duplicates with the same address and name
duplicates_name_address = cafe_data[['location_name', 'address']].duplicated().sum()
print('number of duplicated addresses with the same restaurant name: {}'.format(duplicates_name_address))

number of duplicated addresses with the same restaurant name: 0

Reviewer's comment You are absolutely right. Another option is to exclude the **id** column and look for full duplicates again. Even if you are not going to delete them, it is important to understand what we are working with

Student's comment: This is a good idea, I will evaluate necessary data better in the future

# determine if there are duplicate addresses
duplicate_address = cafe_data[cafe_data['address'].duplicated()]
duplicate_address.sort_values(by = 'address').head(20)

	location_id	location_name	address	chain	location_type	num_seats
5624	17410	SOUPLANTATION # 17	100 N LA CIENEGA BLVD # E-3	True	Restaurant	199
8201	19987	HOME TURF SPORTS BAR	100 WORLD WAY	False	Bar	32
6277	18063	GORDEN BIERSCH	100 WORLD WAY	False	Restaurant	17
6515	18301	STARBUCKS COFFEE	100 WORLD WAY	True	Cafe	9
6000	17786	PINKBERRY #78	100 WORLD WAY	True	Restaurant	3
5996	17782	TORN BASIL	100 WORLD WAY	False	Restaurant	5
8946	20732	ROCCO'S TAVERN	1000 GAYLEY AVE	True	Restaurant	134
6738	18524	LA DODGERS CLUBHOUSE	1000 VIN SCULLY AVE	False	Restaurant	46
3677	15463	LOBBY CAFE BAR	1000 WILSHIRE BLVD	False	Bar	48
7029	18815	CITY CAFE REMOTE STORAGE	1000 WILSHIRE BLVD	False	Restaurant	16
7026	18812	CITY CAFE	1000 WILSHIRE BLVD	False	Restaurant	47
8883	20669	TANDOORI EATS	10004 NATIONAL BLVD	False	Restaurant	7
7036	18822	ORGANIC CORNER CAFE	1001 WILSHIRE BLVD	False	Restaurant	6
9444	21230	WIENERSCHNITZEL #228	1002 S ATLANTIC BLVD	False	Restaurant	4
9591	21377	TACOS Y GUISADOS EL CHARRO	10024 S FIGUEROA ST	True	Fast Food	14
9417	21203	CARLS JR #97	1005 S FAIRFAX AVE	False	Restaurant	94
7634	19420	PORK JANGGUN	1008 S ST ANDREWS PL	False	Restaurant	32
7191	18977	GARLO'S PIES	1010 GLENDON AVE	False	Restaurant	28
7551	19337	LOS ANGELES COUNTRY CLUB	10101 WILSHIRE BLVD	False	Restaurant	4
6206	17992	DRAGON LOCO	1011 W FLORENCE AVE	False	Restaurant	25

• the names paired with the addresses reveal no duplicates
• examination of the duplicate addresses show that there are some locations at the same address
  • a google search on a few of them revealed that they are in a shopping center or mall type space which explains the reason for the same address

# determine the number of duplicate addresses
address_duplicates = cafe_data['address'].duplicated().sum()
print('number of duplicate addresses: {}'.format(address_duplicates))

number of duplicate addresses: 1134

# determine the percent of duplicate addresses
total_entries = len(cafe_data)
percent_duplicate_addresses = (address_duplicates / total_entries)
print('percent of duplicate addresses: {:.2%}'.format(percent_duplicate_addresses))

percent of duplicate addresses: 11.75%

3 Conclusion: Data Preprocessing

Actions Performed

• working version of dataset was created
• null values in chain column were filled with False
  • this value was determine appropriate when examining the original data in Section 2
    • there was only one location found for each
• datatype storage was optimized
• retitled columns to be more readable / user friendly
• looked for missing values
  • none were found
• looked for duplicate rows
  • none were found
• looked for duplicate entries of restarants based on name and address
  • none were found
• looked for addresses that have mulitple restaraunts
  • 1134 restaurants share an address with a different restaurant
  • this accounts for 11.75% of entries
  • a google search on a few of them revealed that they are in a shopping center or mall type space which explains the reason for the same address

Reviewer's comment Very good. Now we can move on to the next step.

4 Analyse the Data

• 4.1 Investigate the proportions of the various types of establishments. Plot a graph.
• 4.2 Investigate the proportions of chain and nonchain establishments. Plot a graph.
• 4.3 Which type of establishment is typically a chain?
• 4.4 What characterizes chains: many establishments with a small number of seats or a few establishments with a lot of seats?
• 4.5 Determine the average number of seats for each type of restaurant. On average, which type of restaurant has the greatest number of seats? Plot graphs.
• 4.6 Put the data on street names from the address column in a separate column.
• 4.7 Plot a graph of the top ten streets by number of restaurants.
• 4.8 Find the number of streets that only have one restaurant.
• 4.9 For streets with a lot of restaurants, look at the distribution of the number of seats. What trends can you see?

4.1 Investigate the proportions of the various types of establishments. Plot a graph.

# determine the number of establishments per type
type_count = cafe_data.groupby('location_type').agg({'location_id': 'count'}).reset_index()
type_count.columns = ['location_type', 'num_per_type']
type_count.sort_values(by = 'num_per_type')

	location_type	num_per_type
0	Bakery	283
1	Bar	292
4	Pizza	320
2	Cafe	435
3	Fast Food	1066
5	Restaurant	7255

# plot a graph
plt.figure(figsize = (12, 6))
# orginal line of code
# type_graph = sns.barplot(data = type_count, x = 'location_type', y = 'num_per_type')
# sorted graph
type_graph = sns.barplot(data = type_count, x = 'location_type', y = 'num_per_type', order = type_count.sort_values('num_per_type').location_type)
plt.title('number of locations per restaurant type', size = 15)
plt.xlabel('location type', size = 15)
plt.ylabel('number of locations', size = 15)

for bar in type_graph.patches:
    type_graph.annotate(format(bar.get_height(), '.0f'), 
                xy=(bar.get_x() + bar.get_width() / 2, bar.get_height()), 
                ha = 'center', 
                va = 'center', 
                xytext = (0, 7), 
                textcoords = 'offset points')
    
plt.show()

Reviewer's comment Please, don't forget to sort values :)

Reviewer's comment 2

😊

Student's comment: The graph is sorted

# create a variable to hold the type labels
rest_labels = type_count['location_type']
# create a variable to hold the values
rest_type_count = type_count['num_per_type']

# plot the data
type_pie = go.Figure(data = [go.Pie(labels = rest_labels, values = rest_type_count)])
type_pie.update_layout(title = 'Proportion of Establishment Types')
type_pie.show()

4.1 CONCLUSION: Investigate the proportions of the various types of establishments. Plot a graph.

• There are 6 types of establishments
• 75% are the restaurant type, followed by fast food at 11%, the remaining make up the remaining 14% (cafe, pizza, bar, bakery)
• Cafes consist of 4.51% of the establishments

Reviewer's comment Great visualization!

Reviewer's comment Why, do you think, restaurants are more common? Please, try to explain it.

Student's comment: Restaurants may be the most popular because patrons wish to have a dining experience that lasts a bit longer than what the other types offer

Reviewer's comment 2

Interesting. To be honest, I don't know why restaurants are more common there. Maybe the point is, LA is a very expensive place to live. One of the most expensive in the USA. And restaurant is usually the most expensive establishment. On the other hand, cafes are a type of restaurant. Although, it cannot clearly define the difference and explain the distribution, I suppose, we can say that restaurants more often provide *food for soul* 😊 While cafe is more often just a place to drink coffee and eat something when you are hungry.

4.2 Investigate the proportions of chain and nonchain establishments. Plot a graph.

# calculate counts of chains and non-chains
chains = cafe_data.groupby('chain').agg({'location_id': 'count'}).reset_index()
chains.columns = ['chain', 'count']
chains

	chain	count
0	False	5975
1	True	3676

# create a varaible to hold the labels for the graph 
chain_labels = ['Non-Chain', 'Chain']
# create a variable to hold the values for the graph
chain_type_count = chains['count']

# graph the data
chain_pie = go.Figure(data = [go.Pie(labels = chain_labels, values = chain_type_count)])
chain_pie.update_layout(title = 'Chain vs Non-Chain Establishments')
chain_pie.show()

4.2 CONCLUSION: Investigate the proportions of chain and nonchain establishments. Plot a graph.

• 62% of the establishments are not part of a chain
• 38% of the establishments are part of a chain

Reviewer's comment Do you have any suggestions about this result?

Student's comment: There may be more non-chain establishments because the city has rules against the number or chains in the city limits, the patrons of the city prefer local and/or unique food establishments, or maybe there are incentives given to non-chain or small businesses, so there are more of them.

Reviewer's comment 2

Yes, it's not so easily to own an establishment chain in such city.

4.3 Which type of establishment is typically a chain?

# review the data
cafe_data.head(2)

	location_id	location_name	address	chain	location_type	num_seats
0	11786	HABITAT COFFEE SHOP	3708 N EAGLE ROCK BLVD	False	Cafe	26
1	11787	REILLY'S	100 WORLD WAY # 120	False	Restaurant	9

# create a table with the number of chains/non-chains per establishment type
type_breakdown = cafe_data.groupby(['location_type', 'chain']).agg({'location_id': 'count'}).reset_index()
type_breakdown.columns = ['location_type', 'chain', 'type_count']
type_breakdown

	location_type	chain	type_count
0	Bakery	False	0
1	Bakery	True	283
2	Bar	False	215
3	Bar	True	77
4	Cafe	False	169
5	Cafe	True	266
6	Fast Food	False	461
7	Fast Food	True	605
8	Pizza	False	167
9	Pizza	True	153
10	Restaurant	False	4963
11	Restaurant	True	2292

# combine the tables
type_breakdown = pd.merge(type_breakdown, type_count, on = 'location_type')

# calculate the chain/non-chain percentages
type_breakdown['percent'] = round((type_breakdown['type_count'] / type_breakdown['num_per_type']) * 100)
type_breakdown

	location_type	chain	type_count	num_per_type	percent
0	Bakery	False	0	283	0.0
1	Bakery	True	283	283	100.0
2	Bar	False	215	292	74.0
3	Bar	True	77	292	26.0
4	Cafe	False	169	435	39.0
5	Cafe	True	266	435	61.0
6	Fast Food	False	461	1066	43.0
7	Fast Food	True	605	1066	57.0
8	Pizza	False	167	320	52.0
9	Pizza	True	153	320	48.0
10	Restaurant	False	4963	7255	68.0
11	Restaurant	True	2292	7255	32.0

# plot the data
plt.figure(figsize = (12, 6))
type_breakdown_graph = sns.barplot(data = type_breakdown, 
                              x = 'location_type', 
                              y ='percent', 
                              hue = 'chain',
                              edgecolor = 'w') #swap y with 'type_count' for counts instead of percent
# plt.title('Establishment Type Breakdown: Chain vs Non-Chain', size = 15)
plt.title('Percentage of Chain and Non-Chain Establishments Per Type', size = 15)

plt.xlabel('establisment type', size = 15)
# plt.ylabel('number of establishments', size = 15)
plt.ylabel('percent of chain and non-chain per type', size = 15)

for bar in type_breakdown_graph.patches:
    type_breakdown_graph.annotate(format(bar.get_height(), '.0f'), 
                xy=(bar.get_x() + bar.get_width() / 2, bar.get_height()),
                ha = 'center', 
                va = 'center', 
                xytext = (0, 7), 
                textcoords = 'offset points')
    
plt.show()

Reviewer's comment Very nice code and graph 👍 The same table can be displayed with a groupby method. Similar to the task about loans, I will use zeros and ones so that the sum is equal to the sum of ones or the number of chain establishments. And count will return the total number of elements in the group. Then the share of chain elements will be calculated as the sum divided by the number. And this is just the average. So I can create this table with just one grouping:

# Reviewer's code

cafe_data.replace({'True': 1, 'False':0}).groupby('location_type')['chain'].agg(['sum','count', lambda x: '{:.2%}'.format(x.mean())]).rename(columns={'sum':'Chain', 'count': 'All Establishments'})

	Chain	All Establishments	<lambda_0>
location_type
Bakery	283	283	100.00%
Bar	77	292	26.37%
Cafe	266	435	61.15%
Fast Food	605	1066	56.75%
Pizza	153	320	47.81%
Restaurant	2292	7255	31.59%

Reviewer's comment It is better to define the legend by ourselves using understandable words like *not chain* and *chain* instead of *false* and *true*.

Student's comment: Thank you for this code example! I am still getting used to know when it is best to use lambda functions. This is much more elegant!

Reviewer's comment 2

Take a look at this [article](https://towardsdatascience.com/lambda-functions-with-practical-examples-in-python-45934f3653a8).

4.3 CONCLUSION: Which type of establishment is typically a chain?

• the breakdown of whether an establishment is a chain or not, is different for the establishment types
• bakeries consist of 100% chains
• the majority of bars and restaurants tend to not be chains
• cafes and fast food have a slightly higher tendency to be chains
• pizzarias appear to be about the same amount of chains and non-chains
• 61% of cafes are chains, 39% are not

Reviewer's comment Interesting. All bakeries are chain establishments.

Reviewer's comment Please, add some explanations. One of the main tasks of any analyst is to explain the results.

Student's comment: The data suggests that establishments that are catering to a grab and go type of patron (bakeries, cafes, fast food) tend to be chains. This could be because they already have a solid business model and are less of a risk to fund vs independent sites. Establishments that encourage more of a sit and stay dining experience tend to be non-chains. For the same reasons listed above as to why there are more non-chains vs chains: - There may be more non-chain establishments because the city has rules against the number or chains in the city limits - the patrons of the city prefer local and/or unique food establishments - maybe there are incentives given to non-chain or small businesses, so there are more of them - They may also be less risky to fund

Reviewer's comment 2

Good :)

4.4 What characterizes chains: many establishments with a small number of seats or a few establishments with a lot of seats?

# review data
cafe_data.head(2)

	location_id	location_name	address	chain	location_type	num_seats
0	11786	HABITAT COFFEE SHOP	3708 N EAGLE ROCK BLVD	False	Cafe	26
1	11787	REILLY'S	100 WORLD WAY # 120	False	Restaurant	9

# make a table with establishments that are only chains
chains_only = cafe_data.query('chain == True')
chains_only.head(2)

	location_id	location_name	address	chain	location_type	num_seats
8	11794	ABC DONUTS	3027 N SAN FERNANDO RD UNIT 103	True	Fast Food	1
10	11796	EL POLLO LOCO	5319 W SUNSET BLVD	True	Restaurant	38

# examine the number of seats in chains
chains_only['num_seats'].describe()

count    3676.000000
mean       39.694233
std        43.437212
min         1.000000
25%        13.000000
50%        25.000000
75%        44.000000
max       229.000000
Name: num_seats, dtype: float64

# calculate the average number of seats in chains
chains_avg = chains_only['num_seats'].mean()
print('average number of seats in chain establishments: {:.0f}'.format(chains_avg))

average number of seats in chain establishments: 40

# distribution plot
plt.figure(figsize = (12, 6))
sns.distplot(chains_only['num_seats'], bins=50)
plt.title('Distribution of Number of Seats in Chains')
plt.xlabel('number of seats', size = 15)
plt.ylabel('density of establishments', size = 15)
plt.show()

# calculate the number of locations that have specific number of seats
chain_seat_counts = chains_only.groupby('num_seats').agg({'location_id': 'count'}).reset_index()
chain_seat_counts.columns = ['num_seats', 'num_locations']
chain_seat_counts.head()

	num_seats	num_locations
0	1	77
1	2	72
2	3	84
3	4	73
4	5	69

# plot a scatter plot for easier visualization
plt.figure(figsize = (12, 6))
sns.scatterplot(data = chain_seat_counts, x = 'num_seats', y = 'num_locations', label = 'chain')
plt.title('Number of Seats in Chain Establishments', size = 15)
plt.xlabel('number of seats', size = 15)
plt.ylabel('number of establishments', size = 15)
plt.show()

• There are three groups of populations of number of seats
  • group one has about 30 seats or less
  • group two has about 30-50 seats
  • group three has greater than 50

# evaluate those with less than and more than 50 seats
under_fifty = chains_only.query('num_seats <= 50')
print('number of chain establishments with less than 50 seats: {}'.format(len(under_fifty)))
over_fifty = chains_only.query('num_seats > 50')
print('number of chain establishments with greater than 50 seats: {}'.format(len(over_fifty)))
under_fifty_perc = len(under_fifty) / len(chains_only)
print('{:.1%} of chain establiments have 50 seats or less'.format(under_fifty_perc))

number of chain establishments with less than 50 seats: 3033
number of chain establishments with greater than 50 seats: 643
82.5% of chain establiments have 50 seats or less

# plot the distribution of 50 seats or less
plt.figure(figsize = (12, 6))
sns.distplot(under_fifty['num_seats'])
plt.title('Distribution of Chains With Less Than 50 Seats', size = 15)
plt.xlabel('number of seats', size = 15)
plt.ylabel('density of establishments', size = 15)
plt.show()

4.4 CONCLUSION: What characterizes chains: many establishments with a small number of seats or a few establishments with a lot of seats?

• Most chains have 50 seats or less
  • There are 3033 chain establishments with 50 seats or less and 643 with greater than 50 seats
    • this is roughly 82.5% and 17.5% respectively
• Those with 50 seats or less have two populations:
  • most have 30 seats or less
• This data set is suggesting that having many establishments with small number of seats is a typical characterization of a chain establishment

Reviewer's comment Yes. A quarter of establishments have no more than 13 seats.

4.5 Determine the average number of seats for each type of restaurant. On average, which type of restaurant has the greatest number of seats? Plot graphs.

# review data
cafe_data.head(2)

	location_id	location_name	address	chain	location_type	num_seats
0	11786	HABITAT COFFEE SHOP	3708 N EAGLE ROCK BLVD	False	Cafe	26
1	11787	REILLY'S	100 WORLD WAY # 120	False	Restaurant	9

# calculate the average number of seats per establishment type
avg_seats = cafe_data.pivot_table(index = 'location_type', values = 'num_seats', aggfunc = 'mean').reset_index()
avg_seats.columns = ['location_type', 'avg_num_seats']
avg_seats['avg_num_seats'] = avg_seats['avg_num_seats'].round()
avg_seats.sort_values(by = 'avg_num_seats')

	location_type	avg_num_seats
0	Bakery	22.0
2	Cafe	25.0
4	Pizza	28.0
3	Fast Food	32.0
1	Bar	45.0
5	Restaurant	48.0

# plot the data
plt.figure(figsize = (12, 6))
avg_seats_graph = sns.barplot(data = avg_seats, x = 'location_type', y = 'avg_num_seats', order = avg_seats.sort_values('avg_num_seats').location_type)
plt.title('Average Number of Seats per Establishment Type', size = 15)
plt.xlabel('establishment type', size = 15)
plt.ylabel('average number of seats')
avg_seats_graph

for bar in avg_seats_graph.patches:
    avg_seats_graph.annotate(format(bar.get_height(), '.0f'), 
                xy=(bar.get_x() + bar.get_width() / 2, bar.get_height()),
                ha = 'center', 
                va = 'center', 
                xytext = (0, 7), 
                textcoords = 'offset points')

plt.show()

Reviewer's comment Great barplot, but again, we should remember about sorting :)

Student's comment: The graph is sorted

4.5 CONCLUSION: Determine the average number of seats for each type of restaurant. On average, which type of restaurant has the greatest number of seats? Plot graphs.

• On average, restaurant types have the greatest number of seats of 48, followed by bars with 45
• Cafes on average have 25 seats

4.6 Put the data on street names from the address column in a separate column.

# review data
cafe_data.head(2)

	location_id	location_name	address	chain	location_type	num_seats
0	11786	HABITAT COFFEE SHOP	3708 N EAGLE ROCK BLVD	False	Cafe	26
1	11787	REILLY'S	100 WORLD WAY # 120	False	Restaurant	9

# separate street number from street
cafe_data[['add_num', 'street']] = cafe_data['address'].str.split(n=1, expand = True)
cafe_data.head()

	location_id	location_name	address	chain	location_type	num_seats	add_num	street
0	11786	HABITAT COFFEE SHOP	3708 N EAGLE ROCK BLVD	False	Cafe	26	3708	N EAGLE ROCK BLVD
1	11787	REILLY'S	100 WORLD WAY # 120	False	Restaurant	9	100	WORLD WAY # 120
2	11788	STREET CHURROS	6801 HOLLYWOOD BLVD # 253	False	Fast Food	20	6801	HOLLYWOOD BLVD # 253
3	11789	TRINITI ECHO PARK	1814 W SUNSET BLVD	False	Restaurant	22	1814	W SUNSET BLVD
4	11790	POLLEN	2100 ECHO PARK AVE	False	Restaurant	20	2100	ECHO PARK AVE

# remove any numbers at the end of the street
cafe_data['street'] = cafe_data['street'].str.rstrip('# 0123456789')
cafe_data.head()

	location_id	location_name	address	chain	location_type	num_seats	add_num	street
0	11786	HABITAT COFFEE SHOP	3708 N EAGLE ROCK BLVD	False	Cafe	26	3708	N EAGLE ROCK BLVD
1	11787	REILLY'S	100 WORLD WAY # 120	False	Restaurant	9	100	WORLD WAY
2	11788	STREET CHURROS	6801 HOLLYWOOD BLVD # 253	False	Fast Food	20	6801	HOLLYWOOD BLVD
3	11789	TRINITI ECHO PARK	1814 W SUNSET BLVD	False	Restaurant	22	1814	W SUNSET BLVD
4	11790	POLLEN	2100 ECHO PARK AVE	False	Restaurant	20	2100	ECHO PARK AVE

# remove 1/2 labels from street
cafe_data['street'] = cafe_data['street'].str.lstrip('"1/2"')
cafe_data.head()

	location_id	location_name	address	chain	location_type	num_seats	add_num	street
0	11786	HABITAT COFFEE SHOP	3708 N EAGLE ROCK BLVD	False	Cafe	26	3708	N EAGLE ROCK BLVD
1	11787	REILLY'S	100 WORLD WAY # 120	False	Restaurant	9	100	WORLD WAY
2	11788	STREET CHURROS	6801 HOLLYWOOD BLVD # 253	False	Fast Food	20	6801	HOLLYWOOD BLVD
3	11789	TRINITI ECHO PARK	1814 W SUNSET BLVD	False	Restaurant	22	1814	W SUNSET BLVD
4	11790	POLLEN	2100 ECHO PARK AVE	False	Restaurant	20	2100	ECHO PARK AVE

4.7 Plot a graph of the top ten streets by number of restaurants.

# count the number of locations per street
street_counts = cafe_data.pivot_table(index = 'street', values = 'location_id', aggfunc = 'count').reset_index()
street_counts.columns = ['street', 'num_per_street']
street_counts.head()

	street	num_per_street
0	3RD ST	1
1	7TH ST	1
2	BEVERLY BLVD	3
3	CITY TERRACE DR	1
4	COLORADO BLVD	1

# create a table of the top 10 streets with the most locations
top_ten = street_counts.sort_values('num_per_street', ascending = False).reset_index()
top_ten_table = top_ten.head(10)
top_ten_table

	index	street	num_per_street
0	1635	W SUNSET BLVD	305
1	1601	W PICO BLVD	305
2	1733	WILSHIRE BLVD	223
3	427	HOLLYWOOD BLVD	194
4	1236	SANTA MONICA BLVD	191
5	1185	S WESTERN AVE	183
6	913	S FIGUEROA ST	173
7	1378	W 3RD ST	173
8	1159	S VERMONT AVE	165
9	1581	W OLYMPIC BLVD	155

# plot the data
plt.figure(figsize = (12, 6))
top_ten_graph = sns.barplot(data = top_ten_table, x = 'street', y = 'num_per_street')
plt.title('Top 10: Streets With The Most Establishments', size = 15)
plt.xlabel('street', size = 15)
# remove this code and replace with plt.xticks(rotation = 30) if add the order parameter to the sns.barplot(... order = ...)
top_ten_graph.set_xticklabels(labels = top_ten_table['street'], rotation = 30)
plt.ylabel('number of locations per street', size = 15)

for bar in top_ten_graph.patches:
    top_ten_graph.annotate(format(bar.get_height(), '.0f'), 
                xy=(bar.get_x() + bar.get_width() / 2, bar.get_height()),
                ha = 'center', 
                va = 'center', 
                xytext = (0, 7), 
                textcoords = 'offset points')
plt.show()

Reviewer's comment You can make it hotizontal so it's easy to read names. Or, for example, you can rotate the labels on the X axis with the `plt.xticks(rotation = )` parameter.

Student's comment: The graph is sorted

4.7 CONCLUSION: Plot a graph of the top ten streets by number of restaurants.

• The top 10 streets with the most restaurants range from 155 - 305 establishments
• The top 2 are W Pico Blvd and W Sunset Blvd, followed by Wilshire Blvd

Reviewer's comment I suppose, these streets are big and crowded.

4.8 Find the number of streets that only have one restaurant.

# calculate the number of streets that only have one restaurant
one_per_street = len(street_counts.query('num_per_street == 1'))
print(one_per_street, 'streets have only one restaurant')

1213 streets have only one restaurant

4.8 CONCLUSION: Find the number of streets that only have one restaurant.

• 1213 streets have only one restaurant

4.9 For streets with a lot of restaurants, look at the distribution of the number of seats. What trends can you see?

# review data
cafe_data.head(2)

	location_id	location_name	address	chain	location_type	num_seats	add_num	street
0	11786	HABITAT COFFEE SHOP	3708 N EAGLE ROCK BLVD	False	Cafe	26	3708	N EAGLE ROCK BLVD
1	11787	REILLY'S	100 WORLD WAY # 120	False	Restaurant	9	100	WORLD WAY

# make a list of top 10 streets
top_ten_list = list(top_ten_table['street'])
top_ten_list

['W SUNSET BLVD',
 'W PICO BLVD',
 'WILSHIRE BLVD',
 'HOLLYWOOD BLVD',
 'SANTA MONICA BLVD',
 'S WESTERN AVE',
 'S FIGUEROA ST',
 'W 3RD ST',
 'S VERMONT AVE',
 'W OLYMPIC BLVD']

# obtain the data on the top 10 streets
top_ten_data = cafe_data.query('street in @top_ten_list')
top_ten_data.head()

	location_id	location_name	address	chain	location_type	num_seats	add_num	street
2	11788	STREET CHURROS	6801 HOLLYWOOD BLVD # 253	False	Fast Food	20	6801	HOLLYWOOD BLVD
3	11789	TRINITI ECHO PARK	1814 W SUNSET BLVD	False	Restaurant	22	1814	W SUNSET BLVD
10	11796	EL POLLO LOCO	5319 W SUNSET BLVD	True	Restaurant	38	5319	W SUNSET BLVD
16	11802	HMS BOUNTY	3357 WILSHIRE BLVD	False	Restaurant	147	3357	WILSHIRE BLVD
22	11808	CAMY'S GRILL	3339 WILSHIRE BLVD	False	Restaurant	39	3339	WILSHIRE BLVD

# calculate the number of establishments per street
top_ten_counts = top_ten_data.pivot_table(index = 'street', values = 'location_id', aggfunc = 'count').reset_index()
top_ten_counts.columns = ['street', 'num_rest_per_street']
top_ten_counts

	street	num_rest_per_street
0	HOLLYWOOD BLVD	194
1	S FIGUEROA ST	173
2	S VERMONT AVE	165
3	S WESTERN AVE	183
4	SANTA MONICA BLVD	191
5	W 3RD ST	173
6	W OLYMPIC BLVD	155
7	W PICO BLVD	305
8	W SUNSET BLVD	305
9	WILSHIRE BLVD	223

# calculate the number of seats per street
top_ten_seats = top_ten_data.pivot_table(index = 'street', values = 'num_seats', aggfunc = ['sum', 'mean', 'median']).reset_index()
top_ten_seats.columns = ['street', 'total_seats_per_street', 'avg_seats_per_street', 'median_seats_per_street']
top_ten_seats

	street	total_seats_per_street	avg_seats_per_street	median_seats_per_street
0	HOLLYWOOD BLVD	10808	55.711340	36
1	S FIGUEROA ST	8276	47.838150	26
2	S VERMONT AVE	7480	45.333333	28
3	S WESTERN AVE	7745	42.322404	28
4	SANTA MONICA BLVD	6840	35.811518	22
5	W 3RD ST	7062	40.820809	28
6	W OLYMPIC BLVD	8199	52.896774	28
7	W PICO BLVD	12342	40.465574	26
8	W SUNSET BLVD	15494	50.800000	35
9	WILSHIRE BLVD	12515	56.121076	37

# create a table of seats and establishments per street
top_ten_table = pd.merge(top_ten_counts, top_ten_seats, on = 'street')
top_ten_table

	street	num_rest_per_street	total_seats_per_street	avg_seats_per_street	median_seats_per_street
0	HOLLYWOOD BLVD	194	10808	55.711340	36
1	S FIGUEROA ST	173	8276	47.838150	26
2	S VERMONT AVE	165	7480	45.333333	28
3	S WESTERN AVE	183	7745	42.322404	28
4	SANTA MONICA BLVD	191	6840	35.811518	22
5	W 3RD ST	173	7062	40.820809	28
6	W OLYMPIC BLVD	155	8199	52.896774	28
7	W PICO BLVD	305	12342	40.465574	26
8	W SUNSET BLVD	305	15494	50.800000	35
9	WILSHIRE BLVD	223	12515	56.121076	37

# plot the data
print('Correlation Between Number of Seats and Number of Establishments per Street')
# plt.figure(figsize = (12, 6))
top_ten_graph = sns.jointplot(data = top_ten_table, x = 'total_seats_per_street', y = 'num_rest_per_street', kind = 'reg')
# plt.title('Correlation Between Number of Seats and Number of Establishments per Street')
plt.xlabel('seats per street', size = 15)
plt.ylabel('establishments per street', size = 15)
top_ten_graph
plt.show()

Correlation Between Number of Seats and Number of Establishments per Street

# distribution of the number of seats on top 10 streets
plt.figure(figsize = (12, 6))
sns.distplot(top_ten_data['num_seats'])
plt.title('Seat Distribution on Top 10 Streets', size = 15)
plt.xlabel('number of seats', size = 15)
plt.ylabel('density', size = 15)
plt.show()

# under 50, top 10 seat distributuion
plt.figure(figsize = (12, 6))
under_fifty_top_ten = top_ten_data.query('num_seats <= 50')
sns.distplot(under_fifty_top_ten['num_seats']) 
plt.title('Seat Distribution on Top 10 Streets with 50 Seats or Less')
plt.xlabel('number of seats', size = 15)
plt.ylabel('density', size = 15)
plt.show()

4.9 CONCLUSION: For streets with a lot of restaurants, look at the distribution of the number of seats. What trends can you see?

• There is a positive correlation between the number of seats and the number or establishments per street, i.e. the more establishments there are, the more seats there are
• Based on the top 10 streets, most streets have between 150 - 200 establishments with around 7300 - 9000 seats
• the distribution of seats for the top 10 streets follows the same pattern for the entire dataset
  • there are 3 populations
    • 0-30 seats, 30-50 seats, 50+ seats
    • most establishments have 0-50 seats

Reviewer's comment Quite obvious result :) Seems like we have the same conclusion about the distributioon.

Reviewer's comment What else can we do here? We can look at the average number of seats for each establishment on these top-10 streets.

Student's comment: Calculated the average # of seats the average number of seats on the top 10 streets is 47. This is higher than the cafe average of 25 for the whole data set. However the median is 29, which is much lower suggesting that the data is right skewed, i.e. there are large outliers. i also looked at the cafe only data for the top 10 streets, see: Extra research: How many cafes are there on each of the top 10 streets? The average and median number of seats for cafes only on the top 10 streets is 22. Because these values are the same, the data is not skewed.

# review the data
top_ten_data.head(2)

	location_id	location_name	address	chain	location_type	num_seats	add_num	street
2	11788	STREET CHURROS	6801 HOLLYWOOD BLVD # 253	False	Fast Food	20	6801	HOLLYWOOD BLVD
3	11789	TRINITI ECHO PARK	1814 W SUNSET BLVD	False	Restaurant	22	1814	W SUNSET BLVD

top_ten_data['num_seats'].describe()

count    2067.000000
mean       46.812288
std        48.987439
min         1.000000
25%        14.500000
50%        29.000000
75%        48.000000
max       229.000000
Name: num_seats, dtype: float64

Extra research: What is the proportion of establishment types on the top 10 streets?

Reviewer's comment Good question.

# review data
top_ten_data.head(2)

	location_id	location_name	address	chain	location_type	num_seats	add_num	street
2	11788	STREET CHURROS	6801 HOLLYWOOD BLVD # 253	False	Fast Food	20	6801	HOLLYWOOD BLVD
3	11789	TRINITI ECHO PARK	1814 W SUNSET BLVD	False	Restaurant	22	1814	W SUNSET BLVD

# create a table with the number of establishment types in the top 10 streets
top_ten_type_counts = top_ten_data.groupby('location_type').agg({'location_id': 'count'}).reset_index()
top_ten_type_counts.columns = ['location_type', 'num_per_type']
top_ten_type_counts

	location_type	num_per_type
0	Bakery	67
1	Bar	59
2	Cafe	82
3	Fast Food	177
4	Pizza	81
5	Restaurant	1601

# plot the data
# create a variable to hold the location information
top_ten_rest_labels = top_ten_type_counts['location_type']
# create a variable to hold the value information
top_ten_type_count = top_ten_type_counts['num_per_type']

top_ten_type_pie = go.Figure(data = [go.Pie(labels = top_ten_rest_labels, values = top_ten_type_count)])
top_ten_type_pie.update_layout(title = 'Top 10 Proportion of Establishment Types')
top_ten_type_pie.show()

Conclusion

• this follows the same trend for the entire data set
  • about 4% of the establishments are cafes on the top 10 streets

Reviewer's comment I think there's an expensive rent there. What do you think?

Student's comment: Because these streets have other establishments and many have high attraction areas, it is likley that the rent is high. But it also suggests that they do quite well financially if they can maintain that location.

Extra research: What are the proportions of chain and non-chain establishments on the top 10 streets?

top_ten_chain_type = top_ten_data.groupby('chain').agg({'location_id': 'count'}).reset_index()
top_ten_chain_type.columns = ['chain', 'count']
top_ten_chain_type.head(2)

	chain	count
0	False	1331
1	True	736

# create a varaible to hold the labels for the graph 
top_ten_chain_labels = ['Non-Chain', 'Chain']
# create a variable to hold the values for the graph
top_ten_chain_type_count = top_ten_chain_type['count']

# graph the data
top_ten_chain_pie = go.Figure(data = [go.Pie(labels = top_ten_chain_labels, values = top_ten_chain_type_count)])
top_ten_chain_pie.update_layout(title = 'Chain vs Non-Chain Establishments in Top 10 Streets')
top_ten_chain_pie.show()

Conclusion

• 35.6% of establishments are chains on the top 10 streets, 64.4% are not chains
  • this follows the trend of the entire data set

Extra research: How many cafes are there on each of the top 10 streets?

# query for the cafe location type
top_ten_cafes_query = top_ten_data.query('location_type == "Cafe"')
top_ten_cafes_query

	location_id	location_name	address	chain	location_type	num_seats	add_num	street
155	11941	MILLIE'S COFFEE SHOP	3524 W SUNSET BLVD	False	Cafe	22	3524	W SUNSET BLVD
156	11942	ANDANTE COFFEE ROASTERS	2201 W SUNSET BLVD	True	Cafe	4	2201	W SUNSET BLVD
495	12281	STARBUCKS COFFEE #6688	1601 WILSHIRE BLVD	True	Cafe	37	1601	WILSHIRE BLVD
1149	12935	STARBUCKS COFFEE #546	10911 W PICO BLVD	True	Cafe	42	10911	W PICO BLVD
1198	12984	CRAFT COFFEE AND GOOD FOOD CAFE	5636 HOLLYWOOD BLVD	True	Cafe	25	5636	HOLLYWOOD BLVD
...	...	...	...	...	...	...	...	...
9281	21067	STARBUCKS COFFEE #5488	5020 WILSHIRE BLVD	True	Cafe	27	5020	WILSHIRE BLVD
9285	21071	STARBUCKS-17413	3584 S FIGUEROA ST # #3	True	Cafe	31	3584	S FIGUEROA ST
9449	21235	CLOVE COFFEE	5007 W PICO BLVD	False	Cafe	8	5007	W PICO BLVD
9598	21384	GOOD PEOPLE COFFEE CO	11609 SANTA MONICA BLVD	False	Cafe	2	11609	SANTA MONICA BLVD
9635	21421	THE TEA & COFFEE EXCHANGE	6801 HOLLYWOOD BLVD # 120	False	Cafe	4	6801	HOLLYWOOD BLVD

82 rows × 8 columns

top_ten_cafes_query['num_seats'].describe()

count    82.000000
mean     22.097561
std      14.347202
min       2.000000
25%       8.500000
50%      22.000000
75%      33.250000
max      49.000000
Name: num_seats, dtype: float64

# create a table to count the number of cafes per street
top_ten_cafes = top_ten_cafes_query.groupby('street').agg({'location_id': 'count'}).reset_index()
top_ten_cafes.columns = ['street', 'num_cafes']
top_ten_cafes.sort_values(by = 'num_cafes', ascending = False)

	street	num_cafes
8	W SUNSET BLVD	21
7	W PICO BLVD	11
9	WILSHIRE BLVD	11
4	SANTA MONICA BLVD	10
0	HOLLYWOOD BLVD	8
1	S FIGUEROA ST	6
5	W 3RD ST	5
3	S WESTERN AVE	4
6	W OLYMPIC BLVD	4
2	S VERMONT AVE	2

order = avg_seats.sort_values('avg_num_seats').location_type

[top_ten_cafes.sort_values(by = 'num_cafes', ascending = False).street]

[8        W SUNSET BLVD
 7          W PICO BLVD
 9        WILSHIRE BLVD
 4    SANTA MONICA BLVD
 0       HOLLYWOOD BLVD
 1        S FIGUEROA ST
 5             W 3RD ST
 3        S WESTERN AVE
 6       W OLYMPIC BLVD
 2        S VERMONT AVE
 Name: street, dtype: object]

# plot the data
plt.figure(figsize = (12, 6))
top_ten_cafes_graph = sns.barplot(data = top_ten_cafes, x = 'street', y = 'num_cafes', 
                                  order = top_ten_cafes.sort_values(by = 'num_cafes', ascending = False).street)
plt.title('Number of Cafes on Top 10 Streets', size = 15)
plt.xlabel('street', size = 15)
plt.ylabel('number of cafes per street', size = 15)

# this line was overriding the order paramater in sns.barplot(... order = ...)
# top_ten_cafes_graph.set_xticklabels(labels = top_ten_cafes['street'], rotation = 30)

# this allows the xticks to be rotated w/o messing with the labels
plt.xticks(rotation = 30)


for bar in top_ten_cafes_graph.patches:
    top_ten_cafes_graph.annotate(format(bar.get_height(), '.0f'), 
                xy=(bar.get_x() + bar.get_width() / 2, bar.get_height()),
                ha = 'center', 
                va = 'center', 
                xytext = (0, 7), 
                textcoords = 'offset points')
plt.show()

Conclusion

• the top 10 streets have between 2-21 cafes per street
• W Sunset Blvd has 21 cafes, followed by W Pico Blvd, Wilshier Blvd, and Santa Monica Blvd: 11, 11, and 10 respectively
  • While this would account for more competition, there may be a larger need for cafes on these streets, so opening a shop on these streets may be profitable

Reviewer's comment Great.

General Conclusion & Recommendations

• Market research was performed on open-source data on restaurants in LA

• The team is investigating opening a cafe with robot servers with an interest in creating a chain establishment

• Establishment types, chain/non-chain establishments, and number of seats were evaluated

• The top 10 streets with the most establishments were investigated specifically

General Conclusions

• There are 6 types of establishments in this data set
  • 75% are restaurants
  • 4.51% are cafes
• The data consists of chains and non-chain establishments
  • 38% are chains, 62% are not chains
  • 61% of cafes are a part of chains
  • Most chains have 50 seats or less
    • Cafes on average have 25 seats
• The top 10 streets
  • have 155 - 305 establishments per street
    • 35.6% of them are chains
    • Most of them have 50 seats or less
  • 4% of these are cafes (82)
    • each street has between 2 - 21 cafes
  • W Sunset Blvd and W Pico Blvd have the two highest amount of establishments of 305 each, followed by Wilshire Blvd with 223
    • These three streets also have the most cafes on them 21, 11, and 11 repectively

Overall

• There is not a lot of competition for cafes
• Most cafes are chains and have on average 25 seats
• The streets with the most cafes also have the most establishments in general
  • this suggests that there is a high demand for food places on these streets perhaps due to tourism, shopping, or business needs

Recommendations

• According to this data set, cafes do not have a lot of competition (compared with restaurants), so this may be a safe type to begin with
• Focus on locations that have a lot of foot traffic, shopping, or other popular places to visit
  • For example: Google says Sunset Blvd is a 'Legendary road for nightlife and landmarks', suggesting there is a lot of people that visit the area
  • Similarly W Pico and Wilshire Blvds also have major shoping and attraction centers
• Look for or build locations that can accomodate 30 - 50 people
• Most cafes are not a part of a chain, however due to the interesting nature of a cafe with robots, this may be feasible after a pilot with one location

Reviewer's comment Totally agree with you.

Presentation Link - PDF

Presentation: https://drive.google.com/file/d/1pDOHgxA-n9iJ1utBtT-p7JtMx6lBUhUT/view?usp=sharing

Overall conclusion

You've done a really good job! - You've shown great presentation and coding skills. - The data was prepared successfully. - There are a lot graphs. It is good, visualization never hurts. - Moreover, the project has a great conclusion, which is also important.

Please, add some explanations for each result. It's one of the most important parts of any research. As soon as you fix it I will accept the project 😉

Student's comment: I've added comments where you requests and updated at the comments. I hope this is what you were looking for. Thank you!